I would love to self-host a Tildes instance with my domain to create a community in my native language speakers. I saw, from old posts, that it is not an easy process at all. But I really love the concept and the style of Tildes. It is better than Reddit and other Fediverse platforms.
Is there anyone who managed to host Tildes instance?

I would be so glad to have some guidance.

I understand that Tildes implements rate limiting for replies to comments in order to discourage excessive back-and-forth debates or arguments. My current rate limit is one reply every 2 hours. So, if I reply to a comment on one post and then try to reply to a comment on another post, it tells me I have to wait 120 minutes (minus however many minutes since my last comment) until I can comment again.

Is this the normal rate limit? If so, don't people find this... limiting?

Update (2025-04-09 at 08:22 UTC): I was just able to comment twice within ten minutes, so it seems the rate limit has disappeared as mysteriously as it appeared.

The popular online discourse on Large Language Models’ (LLMs’) capabilities is often polarized in a way I find annoying and tiresome.

On one end of the spectrum, there is nearly complete dismissal of LLMs: an LLM is just a slightly fancier version of the autocomplete on your phone’s keyboard, there’s nothing to see here, move on (dot org).

This dismissive perspective overlooks some genuinely interesting novel capabilities of LLMs. For example, I can come up with a new joke and ask ChatGPT to explain why it’s funny or come up with a new reasoning problem and ask ChatGPT to solve it. My phone’s keyboard can’t do that.

On the other end of the spectrum, there are eschatological predictions: human-level or superhuman artificial general intelligence (AGI) will likely be developed within 10 years or even within 5 years, and skepticism toward such predictions is “AI denialism”, analogous to climate change denial. Just listen to the experts!

There are inconvenient facts for this narrative, such as that the majority of AI experts give much more conservative timelines for AGI when asked in surveys and disagree with the idea that scaling up LLMs could lead to AGI.

The ARC Prize is an attempt by prominent AI researcher François Chollet (with help from Mike Knoop, who apparently does AI stuff at Zapier) to introduce some scientific rigour into the conversation. There is a monetary prize for open source AI systems that can perform well on a benchmark called ARC-AGI-2, which recently superseded the ARC-AGI benchmark. (“ARC” stands for “Abstract and Reasoning Corpus”.)

ARC-AGI-2 is not a test of whether an AI is an AGI or not. It’s intended to test whether AI systems are making incremental progress toward AGI. The tasks the AI is asked to complete are colour-coded visual puzzles like you might find in a tricky puzzle game. (Example.) The intention is to design tasks that are easy for humans to solve and hard for AI to solve.

The current frontier AI models score less than 5% on ARC-AGI-2. Humans score 60% on average and 100% of tasks have been solved by at least two humans in two attempts or less.

For me, this helps the conversation about AI capabilities because it gives a rigorous test and quantitative measure to my casual, subjective observations that LLMs routinely fail at tasks that are easy for humans.

François Chollet was impressed when OpenAI’s o3 model scored 75.7% on ARC-AGI (the older version of the benchmark). He emphasizes the concept of “fluid intelligence”, which he seems to define as the ability to adapt to new situations and solve novel problems. Chollet thinks that o3 is the first AI system to demonstrate fluid intelligence, although it’s still a low level of fluid intelligence. (o3 also required thousands of dollars’ worth of computation to achieve this result.)

This is the sort of distinction that can’t be teased out by the polarized popular discourse. It’s the sort of nuanced analysis I’ve been seeking out, but which has been drowned out by extreme positions on LLMs that ignore inconvenient facts.

I would like to see more benchmarks that try to do what AGI-AGI-2 does: find problems that humans can easily solve and frontier AI models can’t solve. These sort of benchmarks can help us measure AGI progress much more usefully than the typical benchmarks, which play to LLMs’ strengths (e.g. massive-scale memorization) and don’t challenge them on their weaknesses (e.g. reasoning).

I long to see AGI within my lifetime. But the super short timeframes given by some people in the AI industry feel to me like they border on mania or psychosis. The discussion is unrigorous, with people pulling numbers out of thin air based on gut feeling.

It’s clear that there are many things humans are good at doing that AI can’t do at all (where the humans vs. AI success rate is ~100% vs. ~0%). It serves no constructive purpose to ignore this truth and it may serve AI research to develop rigorous benchmarks around it.

Such benchmarks will at least improve the quality of discussion around AI capabilities, insofar as people pay attention to them.

Update (2024-04-11 at 19:16 UTC): François Chollet has a new 20-minute talk on YouTube that I recommend. I've watched a few videos of Chollet talking about ARC-AGI or ARC-AGI-2, and this one is beautifully succinct: https://www.youtube.com/watch?v=TWHezX43I-4

As per title this is inspired by my recent problems with a Latitude 7320 notebook.

I can't use my desktop right now and so wanted some cheaper nb for normal usage and eventually settled on this model due to being able to get it at an acceptable ratio of price to age and seeing it as compatible on Ubuntu, not noticing the disclaimer until later.

The problems started right after installing Fedora KDE - the nb was running at absolutely abysmal performance and this problem affects several models.

Running passmark I've got above 2000 on cpu, on Windows I had 11000. The cpu was throttling to 1500Mhz and lower for no reason. Switching a BIOS setting of power management to "ultra performance" got me to twice the score.

Eventually using throttled from github for various Lenovo and Dell models and thermald I was able to get to twice that again, still a fifth less than on Windows. Also the repo has potential of security concerns due to how it works, also potential to just stop working due to them later.

Mainly I'm posting this to just say that there is zero legitimate technical reason why this should happen, it works on Windows and on Dell tampered Ubuntu images. The hw is fine but for some reason someone somewhere decided to artificially limit the hw for whatever reason.

Right now I am still indecided if I should write off the several hours I've spent on this and return the machine to play the dice with some other model.

Edit 5.4.: it turns out I was not using the throttled package correctly and now have roughly equivalent performace in Linux as in Windows up from the 4/5 or so after all the other workarounds. All of the points still apply though. I also heartily recommend s-tui as a nice utility for cpu monitoring and stress test.

I have been listening to the PreHistory Podcast, written and produced by an actual academic archaeologist. I enjoy the rigor and specificity but i’m having trouble finding similar content I like, especially without the promotions and ads and fan service. I know that in the age of social media personalities such dry content is hard to come by.

I particularly enjoy ancient history, but feel free to offer other podcasts that feature people who have mastered their discipline and have found clear, effective, and even entertaining ways of sharing it. Thanks!

I recently had a (mostly) civilized discussion with an older gentleman who mentioned reading a book about how newer generations have not learned how to suffer, and that is the root cause of many evils in the world today. He then expressed a sort of excitement at the thought of self-induced suffering through our supreme leader's terrible economic and geopolitical decisions. It would "make us stronger on the other side."

To which my question was, and is still: "You're a top 2% earner in the most powerful country in the world. You have everything you could ever want or dream of. Why do YOU want to suffer?"

My second question/point was: "What you're describing as people being too comfortable, I'd counter that it's just the advancement of technology and industry -- most of us don't HAVE to suffer by breaking our backs working the fields from sun up to sun down because we have equipment to do that for us. Instead, we can work our desk jobs and play games on our phones."

And my final question/point was: "Why would anyone ever wish suffering upon anyone else? That doesn't seem very biblical."

I'm really struggling to understand the line of thinking that I am hearing from the very top levels of the government all the way down to the working class. The thinking that "we deserve to suffer." In a sense, I feel that it's a sort of disguised retribution or malice, i.e. "I don't want to suffer, but there are a bunch of people I disagree with that do need to suffer."

Please help me understand so I can be better prepared to debate the next person who tries to make this point to me. I'm looking at you, Dad.

What have you been doing lately for your own fitness? Try out any new programs or exercises? Have any questions for others about your training? Want to vent about poor behavior in the gym? Started a new diet or have a new recipe you want to share? Anything else health and wellness related?

Anything goes: foods, software, products, bands, websites, appliances, movies, programming languages, travel destinations, etc.

The point of this topic is twofold:

1. To surface some interesting alternatives that could use more exposure.
2. To highlight some of the issues with the currently popular option(s).

Let us know your best “more people should know about this!” swaps, and sell us on why they’re better than the well-known option.

New Thread

Server host: tildes.nore.gg (Running Java 1.21.4)
Bluemap: https://tildes.nore.gg
Playtime Tracker: https://tildes.nore.gg/playtimes.html
Tildes website extension (shows online status & location): Firefox (Desktop and Android) - Chrome
Verification site: https://verify.tildes.nore.gg
Patreon: https://www.patreon.com/TildesMC

The server operates on a soft whitelist. Anyone can log in and walk around, but you need a Tildes account to gain build access.

New Thread

What have you been watching and reading this week? You don't need to give us a whole essay if you don't want to, but please write something! Feel free to talk about something you saw that was cool, something that was bad, ask for recommendations, or anything else you can think of.

If you want to, feel free to find the thing you're talking about and link to its pages on Anilist, MAL, or any other database you use!